Internal Polls and Mayoral Elections

“Internal polling” claims it has “better methodology”.

An email on behalf of London Mayor candidate Shaun Bailey cites “internal polling”. This email also makes claims about public opinion polling.

In short

Hello (from the inside)

An email from the campaign director for Shaun Bailey asserts:

I was looking through our latest internal polling. And I thought you’d want to see this.

At the start of 2020, voters believed Sadiq Khan was the best candidate for Mayor.

They no longer believe this.

A majority of voters now believe Shaun Bailey will do a better job as Mayor.

Image for post
Image for post
Releasing the data tables and methods would help. (Image: Twitter/Ladbrokes Politics)

Internal polls are surveys conducted on behalf of a ongoing political campaign. In the United States and elsewhere, internal polling results can become public.

Campaigners tend to release these results when they are more favourable. That is a publication bias.

Nate Silver (FiveThirtyEight) looked at internal polling for the 2012 Republican presidential candidate. Across seven states, there was an average five-point bias in favour in Mitt Romney.

Image for post
Image for post
The Republican candidate did not win New Hampshire in 2012. (Image: FiveThirtyEight)

The methods of internal polling are often unknown. In public opinion polls, companies should answer key questions about the survey:

The campaign email alludes to “a better methodology”. It fails to mention what that is:

I know the public polls look different — like they did just before Boris Johnson won Mayor of London.

But our internal polls use a better methodology, and we talk to a more representative sample of people.

We do not know the research company, survey mode, sample size and target population. Nor do we know the question wording, order, and response options.

In public polling, Sadiq Khan (Labour) receives about 50% of vote intentions in the first round. The vote intention share estimates for Shaun Bailey (Conservative) are often under 30%.

Surveys provide estimates, subject to many sources of potential error. Electoral contests can change.

Image for post
Image for post
At present, the London mayoral election is on 6th May 2021. (Image: Wikipedia)

Polling in the 2008 London Mayoral election

We should consider the claim that “the public polls look different” to the result in the 2008 contest. This is likely to be misleading.

In 2008, the polls painted a mixed picture. Some polls showed Ken Livingstone (Labour) ahead. Others showed the Conservative candidate in the lead.

Nine of the final 11 polls estimated both main candidates had second round vote shares between 47% and 53%. Survey research implied a close contest.

Image for post
Image for post
YouGov’s final poll estimated Johnson would win 53% of second round votes. (Image: Wikipedia)

There were consistent differences between polling companies. YouGov polling estimated Johnson was ahead. Ipsos MORI and MRUK Cello estimated modest leads for Livingstone. Survey researchers call these consistent differences house effects.

During the campaign, Ken Livingstone threatened to complain about YouGov. After transfers, Johnson won 53% of second round votes. In the end, YouGov were the most accurate company in that election.

Journalists should treat results of internal polling with much suspicion. Information about methods that would help us read these polls are often unreleased.

This blog looks at the use of statistics in Britain and beyond. It is written by RSS Statistical Ambassador and Chartered Statistician @anthonybmasters.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store